You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@mesos.apache.org by Tim St Clair <ts...@redhat.com> on 2013/07/18 16:45:06 UTC

Versioning Scheme?

Hi folks - 

I realize that mesos is in the 0.X cycle and is relatively young, but I thought I would inquire whether folks had thought about a formal versioning schemes?  Those who aren't in the thick of it from day to day can gleen some useful information from formal schemes: 

For example (even[stable] - odd[unstable]) versioning:

--- even [stable] ---
0.14.0 <- Kind of stable (0th rule of software deployment)
0.14.1 <- More stable (usually provoked by bugs found) 
0.14.2 <- When sane IT folks would be willing to install

--- odd [un-stable] ---
0.15.0 <- Bleeding edge, we are lucky it passes the tests ;-) 
0.15.1 <- Ok we actually used it and it seems like jello 
0.15.X <- Tip of master

After some deployment-testing cycle then the swap occurs, and 0.15.Y -> 0.16.0 

This is much akin to the version scheme that we follow with condor, which was essentially borrowed from the kernel.

---------------------------------------------------------------------
My selfish attempt to influence/understand here, is when packagers want to pick versions to place in downstream channels they can pretty easily pick without having to troll through version history.

For example, putting the latest un-stable in Fedora is ok, b/c it's bleeding edge.  On the other hand putting the latest un-stable in EPEL will cause random physicists and IT-admins across the planet to hunt you down and curse your name.   

Thoughts?  
-Tim